Environment Management
Per-service Python virtual environments with package lifecycle and readiness gating
Overview
Each AI service uses its own isolated Python virtual environment. Environments are created on-demand through the admin UI. Each service's overview page shows environment status and handles package installation in phases.
Service Environments
| Service | Config File | Packages File |
|---|---|---|
| LLM | llm_env_config.json | llm_requirements.txt |
| RAG | — | rag_requirements.txt |
| Agents | agents_env_config.json | agents_packages.json |
| MCP | mcp_env_config.json | mcp_packages.json |
| Audio | — | audio_packages.json |
| Doc Extraction | doc_extraction_env_config.json | doc_extraction_packages.json |
| Vision | vision_env_config.json | — |
| Text-to-Image | text_to_image_env_config.json | text_to_image_packages.json |
| Tooling | tooling_env_config.json | tooling_packages.json |
| Studio | — | studio_packages.json |
Phased Overview
Service overview pages progress through three phases:
- Phase A — Environment not created. Shows creation UI only.
- Phase B — Environment exists but required packages missing. Shows install UI.
- Phase C — Environment ready. Shows primary operational view (list/registry or hub).